damage in cement-based materials-resistivity

8/12/2019 Damage in Cement-based Materials-resistivity

1/40

Damage in cement-based materials, studied by electricalresistance measurement

D.D.L. Chung*

Composite Materials Research Laboratory, University at Buffalo, State University of New York, Buffalo,

NY 14260-4400, USA

Abstract

Real-time monitoring of a structural component gives information on the time and condition at which damage

occurs, thereby facilitating the evaluation of the cause of the damage. Moreover, it provides information once

damage occurs, thus enabling timely repair and hazard mitigation. Real-time monitoring also allows study of the

damage evolution. This paper reviews the use of electrical resistance measurement to monitor damage in cement-

based materials. This method is advantageous in its sensitivity to even minor, microscopic and reversible effects.

Damage can occur within a cement-based material, at the interface between concrete and steel rebar, and at the

interface between old concrete and new concrete (as encountered in the use of new concrete for repair of an old

concrete structure). This paper addresses all three aspects of damage, whether the damage is due to static stress,

dynamic stress, freezethaw cycling, creep or drying shrinkage. Moreover, monitoring during straining at various

rates allows study of the effect of strain rate on the damage evolution.

# 2003 Elsevier B.V. All rights reserved.

Keywords: Cement; Concrete; Electrical resistance; Damage; Monitoring; Sensing

1. Damage monitoring

1.1. Technological need

The structural integrity of the civil infrastructure is essential for the safety, productivity andquality of life of the society. This integrity is often a concern due to the aging of the infrastructure,the occurrence of earthquakes, exposure to wind and ocean waves, soil movement, excessive loading,temperature excursions and terrorism. Thus, there is need for monitoring damage nondestructively,

so that timely repair or retirement of structures takes place.

Damage sensing (i.e. structural health monitoring) is valuable for structures for the purpose ofhazard mitigation. It can be conducted during the damage by acoustic emission detection. It can alsobe conducted after the damage by ultrasonic inspection, liquid penetrant inspection, dynamic

mechanical testing or other techniques. Real-time monitoring gives information on the time, loadcondition or other conditions at which damage occurs, thereby facilitating the evaluation of the causeof the damage. Moreover, real-time monitoring provides information as soon as damage occurs, thusenabling timely repair or other hazard precaution measures.

Materials Science and Engineering R 42 (2003) 140

* Tel.: 1-716-645-2593x2243; fax: 1-716-645-3875.

E-mail address: [email protected] (D.D.L. Chung).

0927-796X/$ see front matter # 2003 Elsevier B.V. All rights reserved.

doi:10.1016/S0927-796X(03)00037-8


2/40

Real-time monitoring allows study of the damage evolution, which refers to how damage

evolves in a damaging process and is the subject of much modeling work [112], due to itsfundamental importance in relation to the science of damage. Limited experimental observation of

the damage evolution has involved the use of acoustic emission [1315], thermoelastic stressanalysis[16] and computer tomography (CT) scanning [17]. In contrast, this paper uses electrical

resistance measurement, which is advantageous in its sensitivity to even minor, microscopic and

reversible effects. The ongoing challenge of the experimentalist is to relate such laboratory results to

the structural integrity problems of the civil infrastructure.

Stress application can generate defects, which may be a form of damage in a material. Stress

application can also heal defects, particularly in the case of the stress being compressive. This

healing is induced by stress[18]and is to be distinguished from healing that is induced by liquids,

chemicals or particles[1929]. On the other hand, stress removal can aggravate defects, particularlyin the case of the stress being compressive and the material being brittle. The generation, healing and

aggravation of defects during dynamic loading are referred to as defect dynamics. The little priorattention on defect dynamics is mainly due to the dynamic nature of defect healing and aggravation.

For example, stress application can cause healing, and subsequent unloading can cancel the healing.

This reversible nature of the healing makes the healing observable only in real time during loading.

On the other hand, defect generation tends to be irreversible upon unloading, so it does not require

observation in real time.

Observation in real time during loading is difficult for microscopy, particularly transmission

electron microscopy, which is the type of microscopy that is most suitable for the observation of

microscopic defects. However, observation in real time during loading can be conveniently

performed by electrical measurement. As defects usually increase the electrical resistivity of a

material, defect generation tends to increase the resistivity whereas defect healing tends to decrease

the resistivity.

The strain rate affects the damage evolution during static stress application, in addition to

affecting the mechanical properties in the case of a viscoelastic material. Real-time monitoring by

electrical resistivity measurement during straining at various rates allows study of the effect of strain

rate on the damage evolution. Similarly, real-time electrical resistivity measurement allows

monitoring of the damage evolution during creep, drying shrinkage, fatigue and freezethaw cycling.

1.2. Experimental methods

Damage in cement-based materials is most commonly studied by destructive mechanical testing

after different amounts of damage. However, this method does not allow the monitoring of the

progress of damage on the same specimen and is not sufficiently sensitive to minor damage. As

different specimens can differ in the flaws, damage evolution is more effectively studied bymonitoring one specimen throughout the process rather than interrupting the process at different

times for different specimens. However, the monitoring of one specimen throughout the process

requires a nondestructive method that is sensitive to minor damage. Electrical resistivity

measurement is effective for damage monitoring, particularly in the regime of minor damage, in

addition to monitoring both defect generation and defect healing in real time.

The experimental method used throughout this paper is dc electrical resistance measurement, as

conducted by using the four-probe method. In this method, the outer two electrical contacts are for

passing current, whereas the inner two electrical contacts are for voltage measurement. Each

electrical contact is made by using silver paint in conjunction with copper wire. The contacts are all

on the specimen surface, rather than being embedded in the specimen. In the case of measuring the

2 D.D.L. Chung / Materials Science and Engineering R 42 (2003) 140


3/40

volume resistance, each electrical contact is all the way around the specimen in a plane

perpendicular to the direction of resistance measurement. The use of the two-probe method is

unreliable, as the resistance associated with the electrical contacts is included in the measured

resistance.

1.3. Scope

Damage sensing should be distinguished from strain sensing, as strain can be reversible and is

not necessarily accompanied by damage. Damage can occur within a cement-based material, at the

interface between concrete and steel rebar, at the interface between old concrete and new concrete

(as encountered in the use of new concrete for repair of an old concrete structure), at the interface

between unbonded concrete elements and at the interface between concrete and its carbon fiber epoxy matrix composite retrofit. This paper addresses all these aspects of damage and is focused on

the use of electrical resistance measurement for sensing damage, whether the damage is due to staticstress, dynamic stress, freezethaw cycling, creep or drying shrinkage.

2. Damage due to stress in a cement-based material

This section covers the damage due to stress in a cement-based material that contain no fiber

admixture and in one that contains an electrically conductive fiber admixture. The fiber enhances the

damage sensing ability. In addition, it covers damage at the interface between concrete and steel

rebar and the interface between new concrete and old concrete.

2.1. Damage in a cement-based material without fibers

This section covers the sensing of damage in cement paste, mortar and concrete, all without

fibers. In addition, it covers the effect of strain rate on the damage evolution.

2.1.1. Cement paste

Fig. 1 [30] shows the fractional change in longitudinal resistivity as well as the longitudinal

strain during repeated compressive loading of plain cement paste at an increasing stress amplitude.

The strain varies linearly with the stress up to the highest stress amplitude. The strain returns to zero

at the end of each cycle of loading. During the first loading, the fractional change in resistivity

increases due to defect generation. During the subsequent unloading, the fractional change in

resistivity continues to increase, due to defect aggravation (such as the opening of the microcracks

generated during prior loading). During the second loading, the resistivity decreases slightly as thestress increases up to the maximum stress of the first cycle (due to defect healing in the sense of

reversibility of the measured resistance, i.e. a transient mechanical effect that is caused by pressure

forcing contact to occur after a possibly chemical bond is broken through a brittle failure) and then

increases as the stress increases beyond this value (due to additional defect generation). During

unloading in the second cycle, the resistivity increases significantly (due to defect aggravation,

probably the opening of the microcracks). During the third loading, the resistivity essentially does

not change (or decreases very slightly) as the stress increases to the maximum stress of the third

cycle (probably due to the balance between defect generation and defect healing). Subsequent

unloading causes the resistivity to increase very significantly due to defect aggravation (probably the

opening of the microcracks).

D.D.L. Chung / Materials Science and Engineering R 42 (2003) 140 3


4/40

Fig. 2[31]shows the fractional change in transverse resistivity as well as the transverse strain

(positive due to the Poisson effect) during repeated compressive loading at an increasing stress

amplitude. The strain varies linearly with the stress and returns to zero at the end of each cycle of

loading. During the first loading and the first unloading, the resistivity increases due to defect

generation and defect aggravation, respectively, as also shown by the longitudinal resistivity

variation (Fig. 1). During the second loading, the resistivity first increases (due to defect generation)

and then decreases (due to defect healing). During the second unloading, the resistivity increases due

to defect aggravation. During the third loading, the resistivity decreases due to defect healing.

During the third unloading, the resistivity increases due to defect aggravation.

The variations of the resistivity in the longitudinal and transverse directions upon repeated

loading are consistent in showing defect generation (which dominates during the first loading),

defect healing (which dominates during subsequent loading) and defect aggravation (which

dominates during subsequent unloading). The defect aggravation during unloading follows the defect

Fig. 1. Variation of the fractional change in electrical resistivity with time and of the strain (negative for compressivestrain) with time during dynamic compressive loading at increasing stress amplitudes within the elastic regime for plaincement paste at 28 days of curing.

Fig. 2. Variation of the fractional change in transverse resistivity with time and of the transverse strain with time duringdynamic compressive loading at increasing stress amplitudes within the elastic regime for plain cement paste.



5/40

healing during loading, indicating the reversible (not permanent) nature of the healing, which is

induced by compressive stress. The defect aggravation during unloading also follows the defect

generation during loading.

In spite of the Poisson effect, similar behavior was observed in the longitudinal and transverse

resistivities. This means that these defects are essentially non-directional and that the resistivity

variations are real.

Comparison ofFigs. 1 and 2shows that the increase in resistivity with strain during unloading

in the second cycle is clearer and less noisy for the longitudinal resistivity than the transverse

resistivity. This suggests that defect aggravation is more significantly revealed by the longitudinal

resistivity than the transverse resistivity. Hence, the defects are not completely non-directional.

Identification of the defect type has not been made. Microcracks were mentioned earlier, just for

the sake of illustration. The defects may be associated with certain heterogeneities in the cement paste.

Defects affect the mechanical properties. Therefore, mechanical testing (such as modulus

measurement, which is nondestructive) can be used for studying defect dynamics. However, themodulus is not as sensitive to defect dynamics as the electrical resistivity; the relationship between

stress and strain is not affected while the resistivity is affected. The low sensitivity of the modulus to

defect dynamics is consistent with the fact that the deformation is elastic.

Fig. 3(a)[30]shows the fractional change in resistivity along the stress axis as well as the strain

during repeated compressive loading at an increasing stress amplitude for plain cement paste.

Fig. 3(b) shows the corresponding variation of stress and strain during the repeated loading. The

strain varies linearly with the stress up to the highest stress amplitude ( Fig. 3(b)). The strain does not

return to zero at the end of each cycle of loading, indicating plastic deformation. In contrast,Figs. 1

and 2 are concerned with effects of elastic deformation.

The resistivity increases during loading and unloading in every loading cycle (Fig. 3(a)). The

slope of the curve of resistivity versus time (Fig. 3(a)) increases with time, due to the increasing

stress amplitude cycle by cycle (Fig. 3(b)) and the non-linear increase in damage severity as the

stress amplitude increases. The resistivity increase during loading is attributed to damage infliction.

The resistivity increase during unloading is attributed to the opening of microcracks generated

during loading.

Fig. 4 [30] gives the corresponding plots for silica fume cement paste at the same stress

amplitudes asFig. 3. The strain does not return to zero at the end of each loading cycle, as in Fig. 3.

The resistivity variation is similar toFig. 3, except that the resistivity decreases during loading after

the first cycle. The absence of a resistivity increase during loading after the first cycle is attributed to

the lower tendency for damage infliction in the presence of silica fume, which is known to strengthen

cement[3134]. The resistivity decrease during loading after the first cycle is attributed to the partialclosing of microcracks, as expected since the loading is compressive. In the absence of silica fume (i.e.

plain cement paste,Fig. 3), the effect of damage infliction overshadows that of microcrack closing.Fig. 5[30]gives the corresponding plots for latex cement paste. The resistivity effects are similar

to those ofFig. 4(a), except that the resistivity curve is less noisy and the rate of resistivity increase

during first unloading is higher than that during first loading. This means that the microcrack opening

during unloading has a larger effect on the resistivity than the damage infliction during loading.

Comparison of the results ofFigs. 35for deformation in the plastic regime with those ofFigs. 1and 2for deformation in the elastic regime shows that both the fractional change in resistivity and

the strain are higher in the plastic regime than in the elastic regime by orders of magnitude. Another

difference is that the resistivity decreases are much less significant in the plastic regime than in the

elastic regime. There is no resistivity decrease at all inFig. 3(a), but there are resistivity decreases in

Fig. 1. These differences between the results of plastic and elastic regimes are consistent with the



6/40

much greater damage in plastic deformation than in elastic deformation and the tendency of damage

to increase the resistivity.

That the resistivity decreases are not significant in the plastic deformation regime simplifies theuse of the electrical resistivity to indicate damage. Nevertheless, even when the resistivity decreases

are significant, the resistivity remains a good indicator of damage, which includes that due to

damage infliction (during loading) and that due to microcrack opening. Microcrack closing, which

causes the resistivity decrease, is a type of partial healing, which diminishes the damage. Hence, the

resistivity indicates both damage and healing effects in real time.

2.1.2. Mortars

Figs. 6 and 7[30]show the variation of the fractional change in resistivity with cycle number

during initial cyclic compression of plain mortar and silica fume mortar, respectively. For both

mortars, the resistivity increases abruptly during the first loading (due to defect generation) and

Fig. 3. Variation of the fractional change in electrical resistivity with time (a), of the stress with time (b), and of the strain(negative for compressive strain) with time ((a) and (b)) during dynamic compressive loading at increasing stressamplitudes for plain cement paste.



7/40

increases further during the first unloading (due to defect aggravation). Moreover, the resistivity

decreases during subsequent loading (due to defect healing) and increases during subsequent

unloading (due to defect aggravation); the effect associated with defect healing is much larger forsilica fume mortar than for plain mortar. In addition, this effect intensifies as stress cycling at

increasing stress amplitudes progresses for both mortars, probably due to the increase in the extent of

minor damage. The increase in damage extent is also indicated by the resistivity baseline increasing

gradually cycle by cycle. In spite of the increase in stress amplitude cycle by cycle, defect healing

dominates over defect generation during loading in all cycles other than the first cycle.

Comparison of plain cement paste behavior (Section 2.1.1) and plain mortar behavior (this

section) shows that the behavior is similar, except that the defect healing (i.e. the resistivity decrease

upon loading other than the first loading) is much more significant in the mortar case. This means

that the sandcement interface in the mortar contributes significantly to the defect dynamics,particularly in relation to defect healing.

Fig. 4. Variation of the fractional change in electrical resistivity with time (a), of the stress with time (b), and of the strain(negative for compressive strain) with time ((a) and (b)) during dynamic compressive loading at increasing stressamplitudes for silica fume cement paste.



8/40

Fig. 5. Variation of the fractional change in electrical resistivity with time (a), of the stress with time (b), and of the strain(negative for compressive strain) with time ((a) and (b)) during dynamic compressive loading at increasing stressamplitudes for latex cement paste.

Fig. 6. Variation of the fractional change in resistivity with cycle number (thick curve) and of the compressive strainwith cycle number (thin curve) during repeated compressive loading at increasing stress amplitudes within the elasticregime for plain mortar.



9/40

Comparison of Figs. 6 and 7 shows that silica fume contributes significantly to the defect

dynamics. The associated defects are presumably at the interface between silica fume and cement,

even though this interface is diffuse due to the pozzolanic nature of silica fume. The defects at this

interface are smaller than those at the sandcement interface, but this interface is large in total areadue to the small size of silica fume compared to sand.

Figs. 810 [30] show the fractional change in resistivity in the stress direction versus cycle

number during cyclic compression at a constant stress amplitude in the elastic regime (in contrast to

the increasing stress amplitude in Figs. 6 and 7). Except for the first cycle, the resistivity decreases

with increasing strain in each cycle and then increases upon subsequent unloading in the same cycle.

As cycling progresses, the baseline resistivity continuously increases, such that the increase is quite

abrupt in the first three cycles (Fig. 8) and that subsequent baseline increase is more gradual. In

addition, as cycling progresses, the amplitude of resistivity decrease within a cycle gradually and

continuously increases (Fig. 8).

The increase in baseline resistivity dominates the first cycle (Fig. 9) and corresponds to

a fractional change in resistivity per longitudinal unit strain of1.1 (negative because the strain

was negative). This negative value suggests that the baseline resistivity increase is due to damage

Fig. 7. Variation of the fractional change in resistivity with cycle number (thick curve) and of the compressive strainwith cycle number (thin curve) during repeated compressive loading at increasing stress amplitudes within the elasticregime for silica fume mortar.

Fig. 8. Fractional change in resistivity and strain vs. compressive stress cycle number for cycles 150 for plain mortar.



10/40

(defect generation). The baseline resistivity increase is irreversible, indicating the irreversibility of

the damage. The incremental increase in damage diminishes as cycling progresses, as shown by the

baseline resistivity increasing more gradually as cycling progresses.

The reversible decrease in resistivity within a stress cycle corresponds to a fractional change in

resistivity per unit strain of0.72 at cycle 50 (Fig. 10). It is attributed to defect healing (reversible)

under the compressive stress. As cycling progresses, the cumulative damage (as indicated by the

baseline resistivity) increases and results in a greater degree of defect healing upon compression

(hence, more decrease in resistivity within a cycle).

Both the baseline resistivity and the amplitude of resistivity decrease within a cycle serve as

indicators of the extent of damage. Measurement of the baseline resistance does not need to be done

in real time during loading, thus simplifying the measurement. However, its use in practice is

complicated by possible shifts in the baseline by environmental, polarization and other factors. On

the other hand, the measurement of the amplitude of resistivity decrease must be done in real time

during loading, but it is not much affected by baseline shifts.

The compressive strength before stress cycling is 54:7 1:7 MPa, and after 100 stress cycles is

53:1 2:1 MPa. The modulus, as shown by the change of strain with stress in each cycle, is not

affected by the cycling. Thus, the damage that occurs during the stress cycling is slight, but is still

detectable by resistivity measurement.





11/40

Comparison of the results ofSection 2.1.1on cement paste with those of this section on mortar

shows that the fractional change in resistivity per unit strain (due to irreversible generation of defects

in the elastic regime) is higher for mortar (1.10) than for cement paste (0.10). Moreover, comparison

shows that mortar is more prone to defect healing (reversible) than cement paste, as expected from

the presence of the interface between fine aggregate and cement in mortar.

2.1.3. Concrete

Fig. 11 [30] shows the fractional change in resistance in the stress direction during repeated

compressive loading at increasing stress amplitudes. The resistance increases during loading and

unloading in cycle 1, decreases during loading in all subsequent cycles and increases during

unloading in all subsequent cycles. The higher the stress amplitude, the greater is the amplitude of

resistance variation within a cycle.

The increase in resistance during loading in cycle 1 is attributed to defect generation; that

during subsequent unloading in cycle 1 is attributed to defect aggravation. In all subsequent cycles,

the decrease in resistance during loading is attributed to defect healing and the increase in resistance

during unloading is attributed to defect aggravation.

The results of this section on concrete are consistent with those ofSection 2.1.2on mortar and

those ofSection 2.1.1on cement paste. The compressive strength is higher for mortar than concrete.

The defect dynamics, as indicated by the fractional change in resistance within a cycle, are more

significant for concrete than mortar. The first healing, as indicated by the resistance decrease during

loading in cycle 2, is much more complete for concrete than mortar. These observations mean thatthe interface between mortar and coarse aggregate contributes to the defect dynamics (particularly

healing), due to the interfacial voids and defects.

Figs. 1214[30]show the fractional change in resistance in the stress direction versus cyclenumber during cyclic compression at a constant stress amplitude. Except for the first cycle, the

resistance decreases with increasing stress in each cycle and then increases upon subsequent

unloading in the same cycle. As cycling progresses, the baseline resistivity gradually and irreversibly

increases (Fig. 12). In addition, as cycling progresses, the amplitude of resistance decrease within a

cycle gradually and continuously increases, especially in cycles 19(Fig. 12).In the first cycle, the resistance increases upon loading and unloading, in contrast to all subsequent

cycles, where the resistance decreases upon loading and increases upon unloading (Fig. 13).

Fig. 11. Fractional change in resistance (solid curve) and stress (dashed curve) vs. time during repeated compressiveloading at increasing stress amplitudes for plain concrete.



12/40

The compressive strength before stress cycling is 16:73 0:86 MPa, and after 40 stress cycles

is 14:24 0:97 MPa. Thus, the damage that occurs during the stress cycling is slight, but is still

detectable by resistance measurement.

The gradual increase in baseline resistance as stress cycling progressed (Fig. 12) is attributed to

irreversible and slight damage. The increase in the amplitude of resistance variation as cycling

progresses (Fig. 12) is attributed to the effect of damage on the extent of defect dynamics. In other

Fig. 12. Fractional change in resistance vs. compressive stress cycle number for cycles 140 for plain concrete.

Fig. 13. Fractional change in resistance (solid curve) and stress (dashed curve) vs. compressive stress cycle number forcycles 16 for plain concrete.

Fig. 14. Fractional change in resistance (solid curve) and stress (dashed curve) vs. compressive stress cycle number forcycles 3540 for plain concrete.



13/40

words, the more is the damage, the greater is the extent of defect healing during loading and the

greater is the extent of defect aggravation during unloading.

The fractional loss in compressive strength after the cycling is greater for concrete than mortar,

as expected from the higher compressive strength of mortar. Nevertheless, the baseline resistance

increase is more significant for mortar than concrete, probably due to the relatively large area of the

interface between cement and fine aggregate and the consequent greater sensitivity of the baseline

resistivity to the quality of the interface between cement and fine aggregate than to the quality of the

interface between mortar and coarse aggregate. In other words, the interface between cement and

fine aggregate dominates the irreversible electrical effects.

2.1.4. Effect of strain rate

The mechanical properties of cement-based material are strain rate sensitive. As for most

materials (whether cement-based or not), the measured strength (whether tensile or compressive)

increases with increasing strain rate [35]. This effect is practically important due to the high strainrate encountered in earthquakes and in impact loading. The effect is less for high strength concrete

than normal concrete[36]and is less at a curing age of 28 days than at an early age[37].The cause

of the effect is not completely understood, although it is related to the effect of strain rate on the

crack propagation[35,3840].Although fracture mechanics [36,41,42], failure analysis [38] and mechanical testing over a

wide range of strain rate[38,43,44]have been used to study the phenomenon and cause of the strain

rate sensitivity of cement-based materials, the current level of understanding is limited. This is partly

because of the experimental difficulty of monitoring the microstructural change during loading.

Observation during loading is in contrast to that after loading. The former gives information on the

damage evolution, whether the latter does not. Work on observation during loading is mainly limited

to determination of the stressstrain relationship during loading. Although this relationship is

important and basic, it does not give microstructural information. The use of a nondestructive real-

time monitoring technique during loading is desirable. Microscopy is commonly used for

microstructural observation, but it is usually not sensitive to subtle microstructural changes in a

cement-based material and is not suitable for real-time monitoring. On the other hand, electrical

resistivity measurement is nondestructive and fast.

Fig. 15[45]shows the fractional change in resistivity in the stress direction versus the strain in

the stress direction during compressive testing up to failure of cement mortar (without fiber) at three

different loading rates. The resistivity increases monotonically with strain and stress, such that the

resistivity increase is most significant when the strain or stress is low compared to the strain or stress

at fracture. Similar curvature of the resistivity curve (Fig. 15) occurs for all three loading rates. At

fracture, the resistivity abruptly increases, as expected.Fig. 16[45]shows the stress versus strain at

different loading rates. The stressstrain curve is a straight line up to failure for any of the loadingrates, indicating the brittleness of the failure. The higher the loading rate, the lower is the fractional

change in resistivity at fracture and the higher is the compressive strength, as shown in Table 1.

Table 1

Effect of strain rate on the compressive properties of mortar (without fiber)

Loading rate(MPa s

1)

Strain rate(10

5s

1)

Strength(MPa)

Modulus(GPa)

Ductility(%)

Fractional change inresistivity at fracture

0.144 5.3 41.4 1.6 1.83 0.17 1.9 0.2 1.78 0.240.216 8.8 43.2 1.0 1.85 0.14 1.8 0.2 1.10 0.130.575 23.3 45.7 2.1 1.93 0.17 1.8 0.3 0.81 0.16



14/40

The modulus and ductility essentially do not vary with the loading rate in the range of loading rate

used, although the modulus slightly increases and the ductility slightly decreases with increasing

loading rate, as expected.

The electrical resistivity is a geometry-independent property of a material. The gradual

resistivity increase observed at any of the loading rates as the stress/strain increases indicates the

occurrence of a continuous microstructural change, which involves the generation of defects that

cause the resistivity to increase. The microstructural change is most significant in the early part of

the loading. At any strain, the extent of microstructural change, as indicated by the fractional change

in resistivity, decreases with increasing loading rate. In addition, the amount of damage at failure, as

indicated by the fractional change in resistivity at failure, decreases with increasing strain rate.

Hence, the loading rate affects not only the failure conditions, but also the damage evolution, all the

Fig. 15. Fractional change in resistivity vs. strain during compressive testing up to failure of mortar (without fiber) atloading rates of: (a) 0.144 MPa s1; (b) 0.216 MPa s1; (c) 0.575 MPa s1.

Fig. 16. Stress vs. strain during compressive testing up to failure of mortar (without fiber) at loading rates of: (a)0.144 MPa s1; (b) 0.216 MPa s1; (c) 0.575 MPa s1.



15/40

way from the early part of the loading. A higher loading rate results in less time for microstructural

changes, thereby leading to less damage build-up. The loading rate will likely also affect the

populations of the various types of defects generated, but investigation of defect types and their

populations requires techniques other than electrical resistance measurement. Examples of other

techniques are environmental scanning tunneling microscopy, small angle neutron scattering,

positron annihilation and synchrotron X-ray scattering. However, none of these techniques has been

used for the monitoring of cement-based materials.

2.2. Sensing damage in a cement-based material containing conductive short fibers

In short fiber-reinforced concrete, the bridging of the cracks by fibers limits the crack height to

values much smaller than those of concretes without fiber reinforcement. For example, the crack

height is less than 1mm in carbon fiber-reinforced mortar after compression to 70% of the

compressive strength, but is about 100 mm in mortar without fibers after compression to 70% of thecorresponding compressive strength (Fig. 6 of[46]). As a result, the regime of minor damage is more

dominant when fibers are present.

Cement reinforced with short carbon fibers is attractive due to its high flexural strength and

toughness and low drying shrinkage, in addition to its strain sensing ability. The strain sensing ability

stems from the effect of strain on the microcrack height and the consequent slight pull-out or push-in

of the fiber that bridges the crack [46]. Fiber pull-out occurs during tensile strain and causes an

increase in the contact electrical resistivity at the fibermatrix interface, thereby increasing the

volume resistivity of the composite. Fiber push-in occurs during compressive strain and causes a

decrease in the volume resistivity of the composite [46].

While reversible changes in electrical resistance upon dynamic loading relates to dynamic

strain, irreversible changes in resistance relate to damage. The resistance of carbon fiber-reinforced

cement mortar decreases irreversibly during the early stage of fatigue (the first 10% or less of the

fatigue life) due to matrix damage resulting from multiple cycles of fiber pull-out and push-in [47].

The matrix damage enhances the chance of adjacent fibers to touch one another, thereby decreasing

the resistivity. Beyond the early stage of fatigue and up to the end of the fatigue life, there is no

irreversible resistance change, other than the abrupt resistance increase at fracture[47]. The absence

of an irreversible change before fracture indicates that the mortar is not a good sensor of its fatigue

damage.

Fatigue damage is to be distinguished from damage under increasing stresses. The former

typically involves stress cycling at a low stress amplitude[47], whereas the latter typically involves

higher stresses. The former tends to occur more gradually than the latter. Thus, the failure to sense

fatigue damage[47]does not suggest failure to sense damage, in general.

The sensing of damage under increasing stresses has been demonstrated in carbon fiber-reinforced concrete[48]and carbon fiber-reinforced mortar [49]. The damage is accompanied by a

partially reversible increase in the electrical resistivity of the concrete. The greater the damage, the

larger is the resistivity increase. As fiber breakage would have resulted in an irreversible resistivity

increase, the damage is probably not due to fiber breakage, but due to partially reversible interface

degradation. The interface could be that between fiber and matrix. Damage was observed within the

elastic regime, even in the absence of a change in modulus.

Carbon fiber-reinforced concrete can monitor both strain and damage simultaneously through

electrical resistance measurement. The resistance decreases upon compressive strain and increases

upon damage. This means that the stress/strain condition (during dynamic loading) under which

damage occurs can be obtained, thus facilitating damage origin identification.



16/40

Fig. 17 [48] shows the fractional change in resistance, strain and stress during repeated

compressive loading of carbon fiber (2 wt.% of cement) reinforced concrete at increasing stress

amplitudes up to 20% of the compressive strength (within the elastic regime) [48]. The strain returnsto zero at the end of each loading cycle. The resistance decreases reversibly upon loading in each

cycle. The higher the stress amplitude, the greater is the extent of resistance decrease. As load

cycling progresses, the resistance at zero load decreases gradually cycle by cycle. In addition, an

extra peak in the resistance curve appears after the first 16 cycles inFig. 17and becomes larger and

larger as cycling progresses. The maximum of the extra peak occurs at the maximum stress of the

cycle.

Fig. 18 [48] shows the fractional change in resistance, strain and stress during repeated

compressive loading at increasing and decreasing stress amplitudes. The highest stress amplitude is

40% of the compressive strength. A group of cycles in which the stress amplitude increased cycle by

cycle and then decreased cycle by cycle back to the initial low stress amplitude is hereby referred to

Fig. 17. Fractional change in resistance (upper curve in (a)), strain (lower curve in (a)) and stress (b) during repeatedcompressive loading of carbon fiber-reinforced concrete at increasing stress amplitudes up to 20% of the compressivestrength.



17/40

as a group.Fig. 18[48]shows the results for two groups, plus the beginning of the third group. The

strain returns to zero at the end of each cycle for any of the stress amplitudes, indicating elastic

behavior. Fig. 19 [48] shows a magnified view of the first half of the first group. The resistance

decreases upon loading in each cycle, as inFig. 17. The extra peak at the maximum stress of a cycle

grows as the stress amplitude increases, as inFig. 17. However, in contrast toFig. 18, the extra peak

quickly becomes quite large, due to the higher maximum stress amplitude inFig. 18than inFig. 17.

InFig. 18, there are two peaks per cycle. The original peak (larger peak) occurs at zero stress, while

the extra peak (smaller peak) occurs at the maximum stress. Hence, during loading from zero stress

within a cycle, the resistance drops and then increases sharply, reaching the maximum resistance of

the extra peak at the maximum stress of the cycle. Upon subsequent unloading, the resistance

Fig. 18. Fractional change in resistance (upper curve in (a)), strain (lower curve in (a)) and stress (b) during repeatedcompressive loading of carbon fiber-reinforced concrete at increasing and decreasing stress amplitudes, the highest ofwhich was 40% of the compressive strength.



18/40

decreases and then increases as unloading continues, reaching the maximum resistance of the

original peak at zero stress. In the part of this group where the stress amplitude decreases cycle by

cycle, the extra peak diminishes and disappears, leaving the original peak as the sole peak. In the part

of the second group where the stress amplitude increases cycle by cycle, the original peak (peak at

zero stress) is the sole peak, except that the extra peak (peak at the maximum stress) returns in a

minor way (more minor than in the first group) as the stress amplitude increases. The extra peak

grows as the stress amplitude increases, but, in the part of the second group in which the stress

amplitude decreases cycle by cycle, it quickly diminishes and vanishes, as in the first group. Within

each group, the amplitude of resistance variation increases as the stress amplitude increases and

decreases as the stress amplitude subsequently decreases. The baseline resistance decreases

gradually from the first group to the second group.

Fig. 18(b) [48] shows similar results for three successive groups with the highest stress

amplitude being 60% of the compressive strength. As the stress amplitude increases, the extra peak

at the maximum stress of a cycle grows to the extent that it is comparable to the original peak at zero

stress. The decrease of the baseline resistance from group to group is negligible, in contrast to

Fig. 18(a). Other features ofFig. 18(a) and (b) are similar.

Fig. 20[48]shows four successive groups and the beginning of the fifth group, with the highest

stress amplitude being more than 90% of the compressive strength. The highest stress amplitude isthe same for each group (Fig. 20(b)), but the highest strain amplitude of a group increases from

group to group as load cycling progresses (Fig. 20(a)). In contrast, the highest strain amplitude of a

group does not change from group to group inFig. 20(b). This means that the modulus decreases as

cycling occurs in Fig. 20(a), whereas the modulus does not change in Fig. 20(b). In Fig. 20, the

resistance increases in every cycle. The extra peak at the maximum stress of a cycle is the sole peak

in each cycle. The original peak at zero stress does not appear at all. In each group, the amplitude of

resistance change in a cycle increases with increasing stress amplitude and subsequently decreases

with decreasing stress amplitude. In each group, the resistance increases abruptly as the maximum

stress amplitude of the group is about to be reached. The baseline resistance increases gradually from

group to group.

Fig. 19. A magnified view of the first 500 s ofFig. 18(a).



19/40

Fig. 21[48]shows the results for loading in which the stress amplitude increases cycle by cycleto a maximum (more than 90% of the compressive strength) and is held at the maximum for

numerous cycles (Fig. 21(b)). The strain amplitude (Fig. 21(a)) increases along with the stress

amplitude, but continues to increase after the stress amplitude has reached its maximum. This

indicates a continuous decrease in modulus after the maximum stress amplitude has been reached.

The resistance increases as the stress increases in each cycle, as in Fig. 20. The baseline resistance

increases significantly cycle by cycle and continues to increase after the stress amplitude has reached

its maximum.

Carbon fiber-reinforced concrete is able to sense its own damage, which occurs under

increasing stress even within the elastic regime. The damage is partially reversible, as indicated by

the partially reversible increase in electrical resistivity observed during cyclic loading at a stress

Fig. 20. Fractional change in resistance (upper curve in (a)), strain (lower curve in (a)) and stress (b) during repeatedcompressive loading of carbon fiber-reinforced concrete at increasing and decreasing stress amplitudes, the highest ofwhich was >90% of the compressive strength.



20/40

amplitude which increases cycle by cycle. In contrast, compressive strain is indicated by a reversible

decrease in resistivity. Upon increasing the stress, the group in which the stress amplitude increasescycle by cycle. This resistance increase indicates the occurrence of damage. Upon decreasing the

stress amplitude, the extra peak does not occur, except for the first two cycles of stress amplitude

decrease. The greater the stress amplitude, the larger and the less reversible is the damage-induced

resistance increase (the extra peak). The resistance starts to increase at a stress higher than that in

prior cycles and continues to increase until the stress reaches the maximum in the cycle, thereby

resulting in the extra peak at the maximum stress of a cycle in the part of a partial irreversibility is

clearly shown inFigs. 20 and 21. If the stress amplitude has been experienced before, the damage-

induced resistance increase (the extra peak) is small, as shown by comparing the result of the second

group with that of the first group (Figs. 18 and 19), unless the extent of damage is large (Figs. 20 and

21). When the damage is extensive (as shown by a modulus decrease), damage-induced resistance

Fig. 21. Fractional change in resistance (upper curve in (a)), strain (lower curve in (a)) and stress (b) during repeatedcompressive loading of carbon fiber-reinforced concrete at increasing stress amplitudes up to >90% of the compressivestrength and then with the stress amplitude fixed at the maximum.



21/40

increase occurs in every cycle (Fig. 20), even at a fixed stress amplitude (Fig. 21) or at a decreasing

stress amplitude (Fig. 20), and it can overshadow the strain-induced resistance decrease ( Figs. 20 and

21). Hence, the damage-induced resistance increase occurs mainly during loading (even within the

elastic regime), particularly at a stress above that in prior cycles, unless the stress amplitude is high

and/or damage is extensive.

At a low stress amplitude, the baseline resistance decreases irreversibly and gradually cycle by

cycle (Figs. 17 and 18). This is the same as the effect [46] attributed to matrix damage and

consequent enhancement of the chance of adjacent fibers to touch one another. At a high stress

amplitude, this baseline resistance decrease is overshadowed by the damage-induced resistance

increase, the occurrence of which cycle by cycle as the stress amplitude increases causes the baseline

resistance to increase irreversibly (Figs. 20 and 21). These two opposing baseline effects cause the

baseline to remain flat at an intermediate stress amplitude (Fig. 21).

The baseline resistance in the regime of major damage (with a decrease in modulus) provides a

measure of the extent of damage (i.e. condition monitoring). This measure works in the loaded orunloaded state. In contrast, the measure using the damage-induced resistance increase works only

during stress increase and indicates the occurrence of damage (whether minor or major) as well as

the extent of damage.

The damage causing the partially reversible damage-induced resistance increase is probably

mainly associated with partially reversible degradation of the fibermatrix interface. Thereversibility rules out fiber fracture as the main type of damage, especially at a low stress

amplitude. At a high stress amplitude, the extent of reversibility diminishes and fiber fracture may

contribute to causing the damage. Fiber fracture can occur during the opening of a crack that is

bridged by a fiber. The fibermatrix interface degradation may be associated with slight fiber pull-out upon slight crack opening for cracks that are bridged by fibers. The severity of the damage-

induced resistance increase supports the involvement of the fibers in the damage mechanism, as the

fibers are much more conducting than the matrix.

In the regime of elastic deformation, the damage does not affect the strain permanently, as

shown by the total reversibility of the strain during cyclic loading (Figs. 1721). Nevertheless,damage occurs during stress increase, as shown by the damage-induced resistance increase. Damage

occurs even in the absence of a change in modulus. Hence, the damage-induced resistance increase is

a sensitive indicator of minor damage (without a change in modulus), in addition to being a sensitive

indicator of major damage (with a decrease in modulus). In contrast, the baseline resistance increase

is an indicator of major damage only.

2.3. Damage at the interface between concrete and steel rebar

Steel reinforced concrete is a widely used structural material. The effectiveness of the steelreinforcement depends on the bond between the steel reinforcing bar (rebar) and the concrete.

Destructive measurement of the shear bond strength by pull-out, push-in and related testing

methods is commonly used to assess the quality of the bond [5064]. Nondestructive methods ofbond assessment are attractive for condition evaluation in the field. They include acoustic [6567]and electrical [68]methods. In particular, measurement of the contact electrical resistivity of the

bond interface has been used to investigate the effects of admixtures, water/cement ratio, curing

age, rebar surface treatment and corrosion on the steelconcrete bond[68]. This electrical methodcan be used to monitor in real time the degradation of the bond during cyclic shear loading [69].

Cyclic loading may lead to fatigue and the damage evolution is of scientific and technological

interest.



22/40

Fig. 22[69]shows the fractional change in contact electrical resistance of the joint between steel

and concrete during cyclic shear loading at a shear stress amplitude of 3.73 MPa. The resistance does

not change much upon stress cycling except for an abrupt increase after 8 31 cycles (the particularcycle depending on the sample), when there is no visual sign of damage, and another abrupt increase

at bond failure, which occurs at cycles 220270 (the particular cycle depending on the sample).

Fig. 23 [69] shows the fractional change in contact electrical resistance during cyclic shear

loading at a shear stress amplitude of 0.75 MPa. The resistance abruptly increases after 150210cycles (depending on the sample), due to bond degradation, which is not visually observable. Bond

failure does not occur up to 400 cycles, at which testing is stopped. The bond strength before any

cyclic shear is 6:68 0:24 MPa, and after the abrupt increase (at the end of 400 cycles in Fig. 23) is

5:54 0:43 MPa. Thus, even though the abrupt increase does not cause visually observable damage,

bond degradation occurs.

Comparison of Figs. 22 and 23shows that a higher stress amplitude causes bond degradation

and bond failure to occur at lower number of cycles, as expected. The abrupt increase in resistance

due to bond degradation (not bond failure) (Figs. 22 and 23) provides a method of monitoring bond

quality nondestructively in real time during dynamic loading. In contrast, bond strength

measurement by mechanical testing is destructive. The bond degradation is attributed to fatigue.

Fig. 22. Variation of the fractional contact resistance change with cycle number during cyclic shear loading at a

shear stress amplitude of 3.73 MPa up to bond failure. The contact resistance is that of the interface between concrete andsteel rebar.

Fig. 23. Variation of the fractional contact resistance change with cycle number during cyclic shear loading at a shearstress amplitude of 0.75 MPa. The test was stopped prior to bond failure. The contact resistance is that of the interfacebetween concrete and steel rebar.



23/40

2.4. Damage at the interface between new concrete and old concrete

The repair of a concrete structure commonly involves the bonding of new concrete to the old

concrete[6576]. Partly due to the drying shrinkage of the new concrete, the quality of the bond islimited. Destructive measurement of the shear bond strength has been previously used to assess the

quality of the bond[77].However, the bond may degrade at stresses below the shear bond strength,

even though the degradation may not be visible. This degradation may occur during static or cyclic

loading. In particular, cyclic loading may lead to fatigue. Such degradation is revealed by

measurement of the contact electrical resistance of the bond interface during cyclic shear loading, as

degradation causes the contact resistance to increase[78].

Measurement of the contact electrical resistance between old and new mortars has also been

previously used to assess the performance of carbon fiber-reinforced mortar as an electrical contact

material for cathodic protection [79]. However, the measurement was not carried out during

mechanical loading.Fig. 24[78]shows the fractional change in contact electrical resistance of the joint between old

and new mortars during cyclic shear loading at a shear stress amplitude of 1.21 MPa. The resistance

does not change upon stress cycling except for an abrupt increase after 16 cycles (the particularcycle depending on the sample), when there is no visual sign of damage, and another abrupt increase

at bond failure, which occurs at cycles 1827 (the particular cycle depending on the sample).The bond strength before the first abrupt increase is 2:87 0:18 MPa, and after the first abrupt

increase is 2:38 0:22 MPa. Thus, even though the first abrupt increase does not cause visually

observable damage, bond degradation occurs.

During static loading, the contact resistance increases monotonically with increasing shear

stress and abruptly increases at bond failure, as shown in Fig. 25 [78]for the case of a specimen

which has not been loaded prior to the measurement. No abrupt increase in resistance occurs during

static loading prior to failure, in contrast to the observation of an abrupt increase prior to fatigue

failure (Fig. 24).


loading at a shear stress amplitude of 0.97 MPa (lower than that ofFig. 24). The resistance shows the

Fig. 24. Variation of the fractional contact resistance change with cycle number during cyclic shear loading of a jointbetween old and new mortars at a shear stress amplitude of 1.21 MPa up to bond failure. Thick curve: fractional change incontact resistance. Thin curve: shear stress. The contact resistance is that of the interface between old and new mortars.



24/40

first abrupt increase after 2248 cycles (the particular cycle depending on the sample), and anotherabrupt increase at bond failure, which occurs after 6992 cycles (the particular cycle depending onthe sample).


loading at a shear stress amplitude of 0.81 MPa (lower than that ofFig. 25). The resistance abruptly

increases after 557690 cycles (depending on the sample), due to bond degradation, which is notvisually observable. Bond failure does not occur up to 1300 cycles, at which testing is stopped.

Comparison ofFigs. 24, 26 and 27shows that a higher stress amplitude causes bond degradation

and bond failure to occur at lower number of cycles, as expected. The abrupt increase in resistance

due to bond degradation (not bond failure) (Figs. 24, 26 and 27) provides a method of monitoring

bond quality nondestructively in real time during dynamic loading. In contrast, bond strength

measurement by mechanical testing is destructive. The bond degradation is attributed to fatigue. This

interpretation is consistent with the absence of an abrupt resistance increase during static loading

prior to failure.

Fig. 25. Variation of the fractional contact resistance change with shear stress during static shear loading of a joint betweenold and new mortars up to failure. The contact resistance is that of the interface between old and new mortars.

Fig. 26. Variation of the fractional contact resistance change with cycle number during cyclic shear loading of a jointbetween old and new mortars at a shear stress amplitude of 0.97 MPa up to bond failure. The contact resistance is that ofthe interface between old and new mortars.



25/40

2.5. Damage at the interface between unbonded concrete elements

Many concrete structures involve the direct contact of one cured concrete element with another,

such that one element exerts static pressure on the other due to gravity. In addition, dynamic pressure

may be exerted by live loads on the structure. An example of such a structure is a bridge involving

slabs supported by columns, with dynamic live loads exerted by vehicles traveling on the bridge.

Another example is a concrete floor in the form of slabs supported by columns, with live loads

exerted by people walking on the floor. The interface between concrete elements that are in pressure

contact is of interest, as it affects the integrity and reliability of the assembly. For example,

deformation at the interface affects the interfacial structure, which can affect the effectiveness of

load transfer between the contacting elements and can affect the durability of the interface to the

environment. Moreover, deformation at the interface can affect the dimensional stability of the

assembly. Of particular concern is how the interface is affected by dynamic loads.

Effective study of the interface between concrete elements that are in pressure contact and under

dynamic loading requires the monitoring of the interface during dynamic loading. Hence, a

nondestructive monitoring technique that provides information in real time during dynamic loading

is desirable. Microscopic examination of the interface viewed at the edge cannot effectively provide

interfacial information, though it can be nondestructive and be in real time. Microscopic examination

of the interface surfaces after separation of the contacting elements can provide microstructuralinformation, but it cannot be performed in real time. Mechanical testing of the interface, say under

shear, can provide interfacial information, but it is destructive (unless the shear strain amplitude is

within the elastic regime) and it cannot be conveniently performed in real time (due to the difficulty

of having simultaneous dynamic compression and dynamic shear). The difficulties and

ineffectiveness associated with these conventional techniques contribute to causing the scarcity of

work on concreteconcrete pressure contacts.

In this section, contact electrical resistance measurement is used to monitor concrete concretepressure contacts in real time during dynamic pressure application. As the surface of concrete is

never perfectly smooth, asperites occur on the surface, thus causing the true contact area at the

interface to be much smaller than the geometric junction area. As a consequence, the local stress at

Fig. 27. Variation of the fractional contact resistance change with cycle number during cyclic shear loading of ajoint between old and new mortars at a shear stress amplitude of 0.81 MPa. The test was stopped prior to bond failure.The contact resistance is that of the interface between old and new mortars.



26/40

the asperites is much higher than the overall stress applied to the junction. The greater the true

contact area, the lower is the contact resistance. Deformation (flattening) of the asperites, as caused

by the high local stress at the asperites, increases the true contact area. Therefore, the interfacial

structure is changed. The contact resistance provides information on the interfacial structure,

particularly in relation to the deformation at the interface. By monitoring the contact resistance in

real time during loading and unloading, the extent, reversibility and loading history dependence of

the deformation at various points of loading and unloading can be investigated, thus providing

information on the structure and dynamic behavior of the interface.

Since concrete is somewhat conductive electrically, the contact resistance of the interface

between contacting concrete elements can be conveniently measured by using the concrete elements

as electrical leadstwo for passing current and two for voltage measurement (i.e. the four-probemethod), as provided by two concrete beams that overlap at 908(Fig. 28). The volume resistance of

each lead is negligible compared to the contact resistance of the junction, so the measured resistance

(i.e. voltage divided by current) is the contact resistance. The contact resistance multiplied by the

junction area gives the contact resistivity, which is independent of the junction area and describes the

structure of the interface.

The data further[79]involves the use of mortar (with fine aggregate but not coarse aggregate)

instead of concrete (with both fine and coarse aggregates). However, the interfacial effects should be

quite similar for mortar and concrete.

Fig. 29shows the variation in resistance and stress during cyclic compressive loading at a stressamplitude of 5.0 MPa. The compressive strength of the mortar used is 64 2 MPa, as determined by

compressive testing of 51 mm 51mm 51 mm (2in: 2 in: 2 in.) cubes. The stressstraincurve is a straight line up to failure. In every cycle, the resistance decreases as the compressive stress

increases, such that the maximum stress corresponds to the minimum resistance and the minimum

stress (zero stress) corresponds to the maximum resistance. The minimum resistance (at the

maximum stress) increases slightly as cycling progresses, but the maximum resistance (at the

minimum or zero stress) decreases with cycling. Due to the asperites at the interface, the local

compressive stress on the asperites is much higher than the overall compressive stress. As a result,

plastic deformation occurs at the asperites, which means that more contact area is created during

cycling. The occurrence of deformation is supported by the cross-head displacement observed within

Fig. 28. Sample configuration for measurement of the contact electrical resistance of the interface between unbondedmortar elements.



27/40

each cycle. The displacement is greatest (i.e. most deformation) at the maximum stress within each

cycle and is not totally reversible. The plastic deformation is why the observed electrical resistance

at the minimum stress (i.e. upon unloading) decreases as cycling progresses. On the other hand, due

to the brittleness of the mortar, the compressive loading probably causes fracture at some of the

asperites, thereby generating debris, which increases the contact resistance. Debris generation is

probably the reason for the slight increase in the contact resistance at the maximum stress as cycling

progresses. After about seven loading cycles, the maximum resistance (at the minimum stress) levels

off, due to the limit of the extent of flattening of the asperites. However, the slight increase of the

minimum resistance (at the maximum stress) persists beyond the first seven cycles, probably due to

the continued generation of debris as cycling progresses.

The stress amplitude inFig. 30is 15 MPa, which is higher than that in Fig. 29.The minimum

resistance (at the maximum stress) increases with cycling more significantly than inFig. 29. This is

probably due to the more significant debris generation at the higher stress amplitude. The maximum

resistance (at the minimum stress) increases in the first four cycles. This is probably due to the effect

of debris generation overshadowing the effect of the flattening of the asperites. After four cycles, the

maximum resistance essentially levels off, probably due to the limit of the extent of debris

generation for this stress amplitude.

Fig. 29. Variation of contact resistance (thick curve) with time and of compressive stress (thin curve) with time during cyclic

compression at a stress amplitude of 5 MPa. The contact resistance is that of the interface between unbonded mortar elements.

Fig. 30. Variation of contact resistance with time and of compressive stress with time during cyclic compression at a stressamplitude of 15 MPa. The contact resistance is that of the interface between unbonded mortar elements.



28/40

The results given earlier mean that, even at a low compressive stress amplitude of 5 MPa, the

structure of a concreteconcrete contact changes during dynamic compression. Thus, the interfacialstructure is dependent on the loading history. The debris generation at the interface may be of

practical concern, as the load transfer between the contacting concrete elements may be affected by

the debris.

2.6. Damage at the interface between concrete and its carbon fiberepoxy matrix composite

retrofit

Continuous fiberpolymer matrix composites are increasingly used to retrofit concretestructure, particularly columns [8092]. The retrofit involves wrapping a fiber sheet around aconcrete column or placing a sheet on the surface of a concrete structure, such that the fiber sheet is

adhered to the underlying concrete using a polymer, most commonly epoxy. This method is effective

for the repair of even quite badly damaged concrete structures. Although the fibers and polymer arevery expensive compared to concrete, the alternative of tearing down and rebuilding the concrete

structure is often even more expensive than the composite retrofit. Both glass fibers and carbon

fibers are used for the composite retrofit. Glass fibers are advantageous for their relatively low cost,

but carbon fibers are advantageous for their high tensile modulus.

The effectiveness of a composite retrofit depends on the quality of the bond between the

composite and the underlying concrete, as good bonding is necessary for load transfer. Peel testing

for bond quality evaluation is destructive [93]. Nondestructive methods to evaluate the bond quality

are valuable. They include acoustic methods, which are not sensitive to small amounts of debonding

or bond degradation [94], and dynamic mechanical testing [95]. This section uses electrical

resistance measurement for nondestructive evaluation of the interface between concrete and its

carbon fiber composite retrofit [96]. The method is effective for studying the effect of debonding

stress on the interface. The concept behind the method is that bond degradation causes the electrical

contact between the carbon fiber composite retrofit and the underlying concrete to degrade. Since

concrete is electrically more conductive than air, the presence of an air pocket at the interface causes

the measured apparent volume resistance of the composite retrofit in a direction in the plane of the

interface to increase. Hence, bond degradation is accompanied by an increase in the apparent

resistance of the composite retrofit. Although the polymer matrix (epoxy) is electrically insulating,

the presence of a thin layer of epoxy at the interface was found to be unable to electrically isolate the

composite retrofit from the underlying concrete.

A 40mm 15 mm sample of carbon fiber sheet (the composite retrofit), with the fibers along

the 40 mm length of the sample, is pressed against a surface of the polished concrete block while the

epoxy resin is at the interface for the purpose of bonding the fiber sheet to the concrete, as illustrated

inFig. 31[96]. The curing of the epoxy resin is carried out at room temperature.Four electrical contacts (AD) are applied at four points along the 40 mm length of the fiber

sheet sample, such that each contact is a strip stretching across the 15 mm width of the sample

(Fig. 31). Each electrical contact is in the form of silver paint in conjunction with copper wire. In the

four-probe method used for dc electrical resistance measurement, two of the electrical contacts (A

and D,Fig. 31) are for passing current, and the remaining two contacts (B and C) are for measuring

voltage. The voltage divided by the current gives the measured resistance, which is the apparent

volume resistance of the fiber sheet between B and C when the sheet is in contact with the concrete

substrate.

Uniaxial compression is applied on a concrete block with fiber sheet on one surface, such tat the

stress is in the fiber direction (Fig. 31), while the electrical resistance is continuously measured.



29/40

Fig. 32shows the fractional change in resistance during cyclic compressive loading at a stress

amplitude of 1.3 MPa. The stress is along the fiber direction. Stress returns to zero at the end of each

cycle. In each cycle, the electrical resistance increases reversibly during compressive loading. This is

attributed to the reversible degradation of the bond between carbon fiber sheet and concrete substrate

during compressive loading. This bond degradation decreases the chance for fibers to touch the

concrete substrate, thereby leading to a resistance increase.

Fig. 31. Sample configuration. The vertical arrow indicates the direction of compressive loading. All dimensions are incm. AD are the four electrical leads emanating from the four electrical contacts (thick horizontal lines), which areattached to the fiber retrofit indicated by vertical parallel lines on the front face of the concrete block. The fiber directionis in the stress direction (vertical).

Fig. 32. The fractional change in resistance for the fiber retrofit on a concrete substrate during cyclic compressive loading.



30/40

As cycling progresses, both the maximum and minimum values of the fractional change in

resistance in a cycle decreases. This is attributed to the irreversible disturbance in the fiber

arrangement during repeated loading and unloading. This disturbance increases the chance for fibers

to touch the concrete substrate, thereby causing the resistance to decrease irreversibly as cycling

progresses.

As shown in Fig. 32, the first cycle exhibits the highest value of the fractional change in

resistance. This is due to the greatest extent of bond degradation taking place during the first cycle.

3. Damage due to freezethaw cycling in a cement-based material

Freezethaw cycling is one of the main causes of degradation of concrete in cold regions. Thedegradation stems from the freezing of the water in the concrete upon cooling, and the thawing upon

subsequent heating. The phase transition is accompanied by dimensional change and internal stresschange. Freezethaw cycling can result in failure.

Research on the freezethaw durability of cement-based materials has been focused on themechanical property degradation (e.g. modulus and strength)[9799],weight change[97,99101],length change [101,102], microstructural change [103] and ultrasonic signature change [101,104]

after different amounts of freezethaw cycling. Relatively little attention has been previously givento monitoring during freezethaw cycling. Techniques previously used for real-time monitoringinclude strain measurement [102] and electrical resistivity measurement [105]. Without real-time

monitoring, the degradation could not be monitored during freezethaw cycling. Therefore, study ofthe damage evolution required testing numerous specimens at different number of freezethawcycles. As different specimens are bound to be a little different in the degree of perfection, the testing

of different specimens gives data scatter which makes it difficult to study the damage evolution. In

order to study the damage evolution on a single specimen during freezethaw cycling, anondestructive and sensitive real-time testing method is necessary.

Electrical resistivity measurement is a nondestructive method. The electrical resistivity of

cement paste decreases reversibly upon heating at temperature above 08C (without freezing or

thawing), due to the existence of an activation energy for electrical conduction [106]. This

phenomenon allows cement paste to function as a thermistor for sensing temperature. Thus,

electrical resistivity measurement allows simultaneous monitoring of both temperature and damage.

A temperature increase causes the resistivity to decrease reversibly, whereas damage causes the

resistivity to increase irreversibly.

Fig. 33[107]shows the fractional change in resistivity and the temperature during fast thermal

cycling (40 min per cycle) of mortar (without fiber) between 20 and 528C. The resistivity

decreases upon heating and increases upon cooling in every cycle, due to the existence of anactivation energy for electrical conduction. The resistivity changes smoothly and similarly above and

below 08C, indicating that the phase transition does not affect the resistivity. Even when the heating

and cooling rates are very low, i.e. 6 h per cycle, the phase transition at 0 8C has only slight effect on

the resistivity (Fig. 34). Although the resistivity changes abruptly at 0 8C, the effects of freezing and

thawing on the resistivity are small compared to the effect of temperature on the resistivity. It is

reasonable that this small effect is only observed when the heating and cooling rates are low.

The resistivity at the end of a heatingcooling cycle is higher than that at the beginning of thecycle (Fig. 33). In other words, the upper envelope of the resistivity variation (corresponding to the

resistivity at 208C) increases cycle by cycle. The lower envelope (corresponding to the resistivity

at 528C) also increases cycle by cycle, but the increase is less significant than that of the upper



31/40

envelope. As a consequence, the amplitude of resistivity variation increases with cycling. This

behavior is attributed to damage, which causes the resistivity to increase irreversibly. That the upper

envelope upshifts more than the lower envelope means that the damage occurs more significantly

upon cooling than upon heating. This is expected since: (i) thermal contraction occurs upon cooling

and the surface of the specimen cools faster than the center of the specimen; and (ii) water expands

upon freezing.

Upon freezethaw failure, the resistivity rises abruptly to essentially infinity, as observed beforethe completion of 15 h of cycling (Fig. 33). This rise occurs at 208C (the coldest point of a cycle),

again indicating that damage during cooling is more significant than that during heating. Prior to

failure, no abrupt resistivity increase was observed. This means that the damage evolution involves

damage accumulating gradually cycle by cycle, until failure occurs.

At a given temperature, the resistivity during heating is slightly lower than that during

subsequent cooling, as shown inFig. 34. The hysteresis becomes more severe as cycling progresses.

The hysteresis is attributed to the damage inflicted during cooling and the association of damage

with a higher resistivity. That damage infliction occurs smoothly throughout cooling from 52 to

208C means that the damage is not due to freezing itself, but is due to thermal contraction and the

fact that the surface cools faster than the center of the specimen.

Fig. 33. The fractional change in resistivity vs. time (thick curve) and the temperature vs. time (thin curve) during fastfreezethaw cycling (40 min per cycle) of mortar (without fiber).

Fig. 34. The fractional change in resistivity vs. temperature in a cycle of slow freezethaw cycling (6 h per cycle) ofmortar (without fiber).



32/40

Fig. 35 shows the fractional change in resistivity and the temperature during fast thermal

cycling (40 min per cycle) between 0 and 528C (i.e. without freezing). The resistivity decreases

reversibly upon heating, due to the existence of an activation energy for electrical conduction. In

contrast to the case of freezethaw cycling at a similar cycling rate (Fig. 33), the lower envelope of

resistivity variation does not shift upon cycling and the upper envelope upshifts only slightly. As the

irreversible increase in resistivity is associated with damage, this means that the damage during

thermal cycling without freezing is negligible compared to that during freezethaw cycling. As aresult, failure does not occur after 15 h of thermal cycling without freezing, but was visually

observed before the end of 15 h of freezethaw cycling (Fig. 33).As mentioned earlier, the damage in Fig. 33 was not due to freezing itself, but was due to

thermal contraction and the fact that the surface cooled faster than the center of the specimen.

Comparison between Figs. 33 and 35 shows that the damage caused by thermal contraction is

significant in the presence of freezing, but negligible in the absence of freezing. In other words,

freezing aggravates the damage that is due to thermal contraction.

The thermal damage observed in Fig. 33 is not related to damage that occurs at elevated

temperatures (up to 528C). This is shown by a separate experiment in which the resistivity was

monitored over time up to 4000 s at a constant temperature of 50 8C. The resistivity was observed to

increase by less than 2%, in contrast to the much larger fractional increase in resistivity (whether the

upper envelope or the lower envelope) inFig. 33.

4. Damage due to creep in a cement-based material

Creep[108,109]is a form of time-dependent plastic deformation that occurs under load, which

is typically fixed during creep testing. Creep affects the dimensions of a component and dimensional

stability is important for many structural components, including concrete slabs and columns.

Although creep during concrete curing [110115] is more significant than that after curing [116

120], creep after curing is relevant to the durability and stability of structures during use. Thus, this

section addresses creep after curing. Creep is more severe at elevated temperatures [121,122], but

this section is limited to creep at room temperature.

Fig. 35. The fractional change in resistivity vs. time (thick curve) and the temperature vs. time (thin curve) duringtemperature cycling of mortar (without fiber) without freezing.



33/40

Research on creep has been focused on the strain during creep [123127], rather than thematerial property variation during creep. The microstructure affects numerous properties, including

mechanical and electrical properties. For the purpose of understanding the microstructural effect of

creep, it is desirable to investigate the property variation during creep. It is further preferred that the

property be measurable nondestructively, so that the same specimen can be monitored throughout the

creep process. This section uses the electrical resistivity as the property for nondestructive

measurement. Prior work that involved monitoring a material property during creep was limited to

the stiffness[128]and the ultrasonic pulse velocity[129].

The creep resistance of cement-based materials is affected by admixtures such as silica fume

[130133], fly ash[130,132,134], slag[132]and steel fibers [135,136]. However, this section doesnot address the effect of admixtures.

Fig. 36[137]shows the fractional change in electrical resistivity in the stress direction during

creep testing at a constant compressive stress (20 MPa, compared to the compressive strength of

51 MPa) for plain cement mortar in the cured state (28 days of curing). The fractional change in

resistivity is essentially equal to the fractional change in resistance due to the small strain

involved.

The resistivity increases as creep progresses except for the initial stage of creep, in which the

resistivity drops slightly. The initial drop in resistivity is slight and does not occur in all the

specimens. It is believed to be due to stress-induced healing of defects ( Section 2.1.1).

The main effect of creep is the increase on resistivity, which is attributed to microstructural

change, which can be viewed as minor damage. The fractional change in resistivity per unit strain is500. This value is about the same for specimens which do not exhibit the initial drop in resistivity.

The fractional change in resistivity per unit strain reflects the extent of creep-induced microstructural

change. The value of 500 is comparable to that for drying shrinkage (Section 5) and is larger than the

value of 250 (or less) for static compressive strain (i.e. instantaneous strain rather than creep strain)

(Fig. 15)[45]. This means that the extent of creep-induced microstructural change is comparable to

that of drying shrinkage-induced microstructural change and is larger than that of stress-induced

microstructural change.

The effect of creep on the resistivity is consistent with the effect of strain rate on the resistivity

(Section 2.1.4); the lower the strain rate, the higher is the fractional change in resistivity at the same

strain (Fig. 15).

Fig. 36. Fractional change in resistivity vs. compressive strain during a month of creep testing of plain cement mortar at a

constant compressive stress.



34/40

5. Damage due to drying shrinkage in a cement-based material

The hydration reaction that occurs during the curing of cement causes shrinkage, called

autogenous shrinkage. In case that the curing is conducted in an open atmosphere, as is usually the

case, additional shrinkage occurs due to the movement of water through the p

damage in cement-based materials-resistivity

Documents